Impacts of AMOC weakening in a warm climate on TC activity: GPI results
To compare the change in TC activity governed by anthropogenic forcing to the change in TC activity governed by a weakened AMOC, we compute the basin-averaged GPI for both simulations during a 26-year historical phase (1980–2005) and during the aforementioned future warming scenario (2075–2100; Fig. 2). GPIs are empirically derived indices combining monthly predictors selected for their importance in TC genesis. We normalize GPI to give a global mean value of 90 annual storms over the early 1980–2005 wk simulation. Comparing storm genesis between the historical and future model simulations suggests a general increase in TC activity due to the net effects of anthropogenic warming (Fig. 2, red bar) in nearly all basins, but especially in the North Atlantic and Western North Pacific basins. Globally, considering the net effects of anthropogenic warming, we observe a projected increase of roughly 30 storms per season, corresponding to a 33% increase in TC activity.
Isolating the effect of the AMOC, we further find that the weakening of the AMOC is the dominant effect in strengthening TC genesis from the fx to wk simulations for the late 21st -century climate (Fig. 2, blue bar) in the North Atlantic and Western North Pacific basins. For example, the effects of AMOC slowdown explain roughly 60% of the projected increase in 12 storms per season in the Atlantic basin in a warmer climate. The effects of AMOC slowdown explain roughly 100% of the storm count increase in the North Indian and Western South Pacific basins in a late 21st century warming scenario (i.e. TC changes in these regions in the fx simulation are nearly zero).
Next, we compare the spatial distribution of GPI during TC active seasons between the wk and fx simulations (Fig. 3a). Firstly, we observe a poleward migration under a warm, weakened AMOC scenario in the North Atlantic in comparison to the fixed AMOC scenario. These results complement previous studies (e.g., refs 39–42) that find that anthropogenic forcing yields a poleward shift of TC activity, but pinpoint the role of AMOC weakening in this shift. We also observe an increase in GPI near the Western coast of Africa, the Northeastern coast of Australia, and near coastlines in the Arabian Sea and the Bay of Bengal in a warmer climate with a weakened AMOC
Perhaps most strikingly, we find that in the North Atlantic basin, a weakened AMOC in a warmer climate yields a strong increase in GPI along the eastern coast of the U.S. and in the Gulf of Mexico, which has significant socio-economic implications for future storm planning in populated regions along the coastal U.S. In contrast, the North Atlantic TC main development region exhibits a slight decrease in GPI.
Figures 3b. and 3c. illustrate the impact of a weakened AMOC on two key variables that constitute the GPI formula (Eq. 1) – potential intensity (PI) and vertical wind shear (also see Extended Data Fig. 1). First, in the North Atlantic with a weakened AMOC, we observe an increase PI above 15°N. A notable increase in PI along the U.S. eastern coastline corresponds to regions of increased GPI in the scenario with a weaker AMOC.
Similarly, in the Indian Ocean, along the northern coastline of Australia, and in the Western North Pacific basin near the coastlines of the Philippines and Vietnam, an increase in PI linked to the weakening of the AMOC plays a key role in the GPI increase. Also, a decrease in PI in the North Atlantic at latitudes below 15°N, in the Southern Indian Ocean, and in the Southern West Pacific (due to AMOC slowdown) likely contributes to the decrease in GPI in these regions.
Other spatial changes in GPI appear to be governed largely by changes in vertical wind shear (Methods). First, in the NA, we observe that changes in GPI in the TC main development region largely resemble the difference in vertical wind shear under a weakened AMOC scenario. We observe a slight decrease in vertical wind shear along the U.S. coastline, suggesting that a weakened AMOC contributes to a decrease in shear. We also find that a weakened AMOC contributes to an increase in Caribbean vertical wind shear and slight decrease in vertical wind shear off the western coast of Africa, which aligns with the results presented by Yan et al. 2017 (33; see Fig. 3). The increase in Caribbean wind shear, along with the decrease in shear near the U.S. coastline, appear to contribute to the respective slight decrease and increase in GPI in these regions of the basin.
Other spatial changes in global GPI connected to a weakened AMOC in a warm climate are influenced by a combination of variables included in the GPI equation (Eq. 1), such as mid-tropospheric moist entropy deficit and absolute vorticity, but their effects are minor (Extended Data Fig. 1). In summary, in a warm climate with AMOC decline, changes in PI and vertical wind shear affect GPI the most.
Mechanisms for the increase in projected TC activity
We next focus on the mechanisms that determine the impacts of AMOC slowdown on potential intensity and vertical wind shear, especially in the NA. Firstly, given the important relationship between PI and SST (43), to understand the connection between a weakening AMOC and increased PI along the US eastern seaboard, we examine the difference in SST between the two model simulations (Fig. 1c). Most notably, a weakening of the AMOC acts to reduce SST in the subpolar North Atlantic, known as the North Atlantic Warming Hole (NAWH), north of 40ºN. However, we observe only a negligible relative decrease in SST in the NA region where PI increases with a weakened AMOC (roughly between 15 and 35º). Thus, changes in SSTs due to AMOC weakening in the Atlantic Ocean should not have a strong effect on PI in this region.
On the other hand, PI critically depends on air-sea thermodynamic disequilibrium and Carnot efficiency (44). Figure 4 illustrates the difference in air-sea thermodynamic disequilibrium and the modified Carnot efficiency (Eq. 5) due to AMOC weakening. With a weaker AMOC, we observe a relative increase in air-sea thermodynamic disequilibrium (mean values of TC-season air-sea thermodynamic disequilibrium and Carnot efficiency are shown in Extended Data Fig. 5). The observed increase in thermodynamic disequilibrium, for the weakened AMOC, is associated with a stronger relative cooling of the troposphere than the ocean in the subtropical North Atlantic in the wk experiment (Extended Data Fig. 4 and Fig. 1). We emphasize that here we describe a relative cooling (the wk minus fx simulation), since the troposphere actually warms in both experiments. Since the Carnot efficiency does not change much for a weakened AMOC in the regions of interest (Fig. 4b), we conclude that a heightened air-sea thermodynamic disequilibrium is the key factor contributing to a higher PI with a weaker AMOC.
Next, given the potential influence of vertical wind shear on the spatial shift of projected TCs and overall basin-wide TC activity when comparing the two simulations, both of which come from one climate model (CCSM4), we further examine how robust the impact of AMOC decline on global vertical wind shear patterns is across a broader selection of climate models. Specifically, we regress vertical windshear onto the change in strength of the AMOC in RCP8.5 and historical CMIP6 model simulations (Fig. 5). Figure 5 confirms that a decrease in AMOC strength is associated with statistically significant changes in vertical wind shear, particularly in the Atlantic. We find that across the simulations, AMOC weakening is associated with a decrease (increase) in vertical wind shear near the U.S. eastern seaboard (Caribbean and TC Main Develop Region), and these results are in general agreement with our CCSM4 simulations. We note that Fig. 5 also reveals that there is large variability in the magnitude of projected AMOC weakening across different GCMs, as well as in wind shear changes. In our particular climate GCM, changes in wind shear due to AMOC slowdown appear to be generally less important than changes in TC potential intensity.
Impacts of AMOC weakening in a warm climate on TC activity: dynamical downscaling
To confirm the inferences from the GPI analysis and provide an independent assessment of the role of AMOC in TC activity, we turn to downscaling TC tracks for the wk and fx simulations for the period 2075–2100 as depicted in Fig. 6. The downscaling model makes use of large-scale variables from the GCM simulations to propagate seeded atmospheric vortices while estimating their evolving amplitude (Methods). A relatively small fraction of imposed “seeds” develops into tropical cyclones and is used in the analysis.
Critically, Fig. 6b reveals a greater TC track density in the North Atlantic in the wk simulation in comparison the fx simulation, which is consistent with increased GPI values along the U.S. coastline is the experiments with a weakened AMOC (Fig. 3). We also observe an increase in track density with a weakened AMOC in the Southern Indian Ocean, Western South Pacific, and in the Western North Pacific near the coast of Japan. These results generally agree with GPI predictions, even though there are some differences (notably in the Western North Pacific and north Australian regions).
Furthermore, we find that AMOC slowdown in a warm climate contributes to TC poleward migration (42) in the North Atlantic since AMOC slowdown appears to trigger more high latitude TC activity in the North Atlantic.